Are robots the key to omnichannel inventory management?
The use of robots on the retail sales floor is the same sort of double-edged sword as when used for the warehouse. Some see them as capable of handling arduous processes to allow employees to focus on customer service. Others see robots as pure replacements, threatening to cut out the need for real live human staff. The latest robot to make an appearance at the front of the store is, if it catches on, likely to generate the same sort of controversy.
A company called 4D Retail Technology Corp. has created a robot capable of automating the inventory process by rolling through the aisles and imaging every product and every barcode in a store. An article on ZDNet described the robot, called the 4D Space Genius, as being a Segway-mounted scanner, which can successfully traverse and scan a full grocery store in about an hour. The robot can identify not only the location of every item in the store, but can log discrepancies between price tags, missing price tags and empty shelves.
Among the robots’ potential applications are determining stock position closer to real-time and streamlining ordering and re-shelving. ZDNet author Greg Nichols noted, for instance, that the ability of the robot to create a 3-D map of a store and its products could theoretically create a closer tie between online and in-store shopping. Demonstration units are slated to appear at a few unnamed major chains in North America and Europe.
Others are also working on using robots to crack the difficult nut of streamlined inventory management. The creators of Lowe’s OSHBot are working to get the robots to track inventory in real-time, according to a recent Fast Company article. The developers also working on building out other inventory-related solutions within the robot, such as the ability to tell a customer if an item s/he brings into the store is stocked there.
Others retailers, such as Target, have been using less flashy RFID technology to more accurately manage inventory and integrate it with omnichannel and online initiatives.
Discussion Questions
DISCUSSION QUESTIONS:
What are the potential upsides and downsides of implementing the 4D Space Genius in grocery stores? Will this be the technology solution for real-time inventory management, or are other options such as RFID a better bet?
Poll
BrainTrust
Max Goldberg
President, Max Goldberg & Associates
Ken Lonyai
Consultant, Strategist, Tech Innovator, UX Evangelist
Assuming this actually works (and there’s no reason to not assume that) and can measure inventory, then it becomes a cost question — is it cheaper to have the robot determine what needs to be re-stocked?
As described, the robots could be a big benefit in tracking inventory and replenishing. Few things frustrate consumers more than out-of-stocks. Target has shown us that RFID, at least how they implement it, doesn’t work to consumers’ satisfaction. Perhaps with a commitment from management, and some of these new robots, Target could take some of the hassle out of shopping in its stores.
There is a larger philosophical issue at play here that goes beyond retail. It’s a matter of how an industrialized society allows automation and technology to affect humans in ways that are potentially detrimental. This issue has been accelerating at an ever quickening pace as the information age has progressed. Software, imaging/tracking technologies and robots are almost always superior to humans at the mundane task of managing things. So from a business perspective, it makes sense to utilize such systems over people. However, as these systems are brought online, at best the roles for humans change, if not get obsoleted. So with limited budgets, any entity, retailer, distributor, manufacturer, etc., has to determine what path to take balancing what’s best for business and what obligations it has to employees.
I believe in the technology and that if not now, in a few iterations it will deliver what’s promised, but I also believe that we need to determine how to address obsoleting jobs as a result of these systems before leaping headfirst into this new territory.
Robots could, in fact, enable other technologies that have been “sputtering” for years, like RFID. Shoppers are more than ready for robots in stores, as evidenced by concept stores internationally. Also, robots will be more effective in stocking physical shelves in stores to truly minimize out-of-stocks. Further, they will manage inventory across channels and locations closer to real-time than humans can.
Robots in stores have been around for years now, and with optical recognition and two-way shopper interaction, the time is now to move beyond the novelty and get serious.
The fastest way for a retailer to grow sales is to be in-stock!
There are some things that a robot can do faster and cheaper than human staff. Inventory management and stocking shelves can definitely be one of the areas where robots can be cost effective. The capability of price checking and and consistency is an added bonus.
One of the rapidly growing components of omnichannel is BOPIS — Buy Online Pickup In Store. The challenge with BOPIS is real-time inventory management to make sure what the customer buys online is in fact on the shelf. Robots are the perfect solution for BOPIS.
There is little question of the capability and potential of robots for to replace some of the staffing required for inventory and stocking. The real question is whether retailers will try to cut staff by using robots, or if they will use robots to free up staff to create the customer experience that will differentiate their stores in ways that will make customers come back.
If I read this correctly, it is an auditing robot which scans and shows what is on the shelves, humans still have to go through and restock it. I can see it as an auditing tool going from store to store to do audits, versus putting one permanently in one store. This would most likely be used after hours so there is less consumer interaction, but i do wonder how well it will navigate the aisles during stocking time when there are pallets in the store.
RFID and shelf technology will in the end beat the bots and the issues that they will bring, from maintenance to lawsuits for ramming into people. RFID does not scare customers’ kids.
Good to test out things like 4D’s bots, but we have simpler technology that can have even more impact than a bot.
We are much closer to RFID on all items than getting Cyberdyne Systems online!
Rather than regurgitate the information above, let’s just say this: This is just the beginning. It is a strange, yet very exciting, time we live in as we see robots moving through aisles of a grocery stores and driver-less cars dropping off and picking up our customers. And the computer that drives this will allow retailers to gain incredible real-time data that will help streamline business, track trends and much, much more.
We really don’t need something else blocking aisle access in stores. Particularly big box stores that often have pallets lying around. Now I am not opposed to having kiosks for way-finding, etc., but these should be off the beaten path.
For my 2 cents, RFID works just fine!
This is a good example of creating technology instead of transforming the paradigm of inventory management. This robot is design to mimic the actions of humans, not optimize the inventory management experience.
The real question is why even have the items on the shelves on the first place? Grab-and-go is fine, but how about being more innovative and having the shelves and robots in a warehouse/separate location picking for customers who can just scan the product and fill up a virtual cart on their mobile phone instead of having robots/humans mingle on the floor together?
Technology like this is novelty and not a solution to the real issue of optimizing inventory management and improve the shopping experience.
Up-sides:
Robots have the potential to dramatically increase the efficiency and accuracy for retailers to identify every product on their shelves, ensure accurate pricing, monitor stock levels, and analyze product geoinformation. Today, these are not favorite tasks for employees and they are not done very frequently.
Therefore, for most retailers, decisions regarding shelf space and product placement are primarily based on old data (yesterdays polling), faulty math (sales, shipments with no shrink information) and broad estimations of what they think is on the shelves. With continuous store scans by robots, retailers will be able to make decisions based on real-time, accurate inventory data.
Another practical use for robots is helping shoppers locate products in the store. For simple requests that the robot understands, it could be a convenient and fun way for shoppers to follow the robot the the product they are looking for.
Down-sides:
While the use of robots in stores present some significant efficiencies, there are a few aspects that I think will be hard for robots to replace humans. While facial recognition has the potential for robots to recognize individual shoppers, it will be difficult for a robot to provide the personal recognition and friendly dialog that sales associates can offer. Another challenge with robots is voice recognition for non-structured requests from shoppers. I hope the store robots will more accurately understand my verbal requests than Siri does on my iPhone.
Despite having been in this industry for a few years, with the technology available today I don’t really understand why we need a robot to check the shelves to see what’s in stock when knowing inventory positions is a relatively simple bit of basic addition and subtraction.
Yes I know we’d have to scan product out of backrooms (which of course is impossible), and connect the POS to our inventory keeping system in real time (a task for which computers have yet to be built that can handle ….) and make some assumptions about product in shopping carts for what, 10 minutes (and not a concern for non grocers). I still fail to understand why a decent real time inventory position can’t be had with less than a robot.
BTW image recognition on shelves isn’t 100% either, some of those liquid filled bottles can be confusing. And if the robot hasn’t scanned a shelf in the past what 30 minutes then it really isn’t real time either.
Hal, how much 2% milk is left in the cooler? (What’s that, can’t see through the frosty glass door, and can’t count how many products are behind the first one?….)
I vote for RFID.
I’ve pointed out that the real conundrum for bricks retailers in responding to online competition, is that their operational business model is built on pallets of merchandise delivered to stores, and unpaid stock-pickers, aka shoppers removing the inventory and handling delivery to themselves, too!
The robot described in this article is sorely needed to just get a handle on the information that is needed, about the MASSIVE breadth of inventory in the store. It is an information only robot, that can reduce the cost of stocking shelves, but does nothing about the cost of PAYING stock pickers to replace the unpaid ones, for example, for click & collect.
This is just one major obstacle to introducing click & collect, that I don’t see anyone having great success with. How long, and how many billions of dollars will be spent, to get click & collect up to the 5-10% usage rate of smartphones as shopping assistants?
Economic/operational issues alone will reduce the success of click & collect, for all but high margin sales that can bear the cost of paying for what the retailer is accustomed to getting in large quantity, FREE! Unpaid stock-pickers = shoppers. Wake up, WAKE UP!!!
(The one possibility would be partnering bricks with Amazon, who already has an ITEM logistic management system in place.)
Great PR coup for the 4D folks, but the ZDNet writer didn’t do much homework on the story. If he had, he’d have discovered that Space Genius is at least the fourth or fifth to try this “robotic” concept in retail stores.
I’ve listed and linked a bunch of similar (and not-so-similar) examples in my January post.
And by the way, who says you need a robotic device to run a bunch of cameras up and down store aisles to capture inventory information? A little outfit called ShelfSnap introduced the concept with cameras mounted on hand-pushed trolleys in 2008. It was focused on merchandising compliance, capturing a photo of a shelf-set that could be compared against a two-dimensional planogram.
The crucial factor, truly, is not whether you can self-propel the camera array through the store environment. It is and has always been the machine-vision capabilities and speed of data transfer.
Today’s super hi-res cameras can certainly read more granular information including codes on shelf tags, or detect voids on the shelf or peg rack. But in an FMCG world, there are 35K-50K instances (SKUs) to survey every reorder cycle, and that means an enormous amount of unstructured data (i.e. images) to capture, process, and react to. And THAT means you need to embed the image analytics on the trolley OR transfer the image data rapidly back to a processor in the cloud for communication to your response system.
Complicating factors include how a machine vision system can count the number of packages on a tightly merchandised shelf and recognize an SKU that has just one or two units remaining near the front of the shelf (an OOS in the making).
Even harder for the machine to identify are locations that have been “faced over” by merchandisers who move adjacent packages to fill in a void. And don’t get me started on products in flexible pouches with shiny surfaces — can the system dependably distinguish among three flavors of the same brand of beef jerky by image-rec alone? What if they are partially turned on the shelf?
Even 99% accuracy for such a brute-force store scanning solution may be inadequate in a retail operation. The remaining 1% of 50,000 items would equal 500 different (and unknown) instances of inventory reordering inaccuracy in each store every night.
Robots see everything but know nothing. Store employees certainly miss a lot, but they know an under-stocked or faced-over item when they see one, and they often know how to fix it.
To be fair, I believe the machine vision problems might well be improved by the use of “learning systems” that get smarter over time. Or, as others here suggest, item-level RFID may help improve counting accuracy for some categories of goods (it’s a long long way from helping on planogram compliance).
Do we really need a master cylinder riding a Segway for that? Maybe just a flock of tiny camera drones that whirr through the store capturing image data, with a few safety sensors built in to avoid collisions….
Oh — and I can see absolutely no connection whatsoever between these devices and in-store order picking for click-and-collect. When each item is scanned as an order is built, the stock on hand is immediately decremented for the store’s inventory reordering system. Robots would be redundant.